Good morning, everyone. Welcome to our next session at the DB Tech Conference. I'm Matt Niknam, communications analyst here at Deutsche Bank. We are very pleased to be joined by Cogent CEO, Dave Schaeffer. Dave, welcome back. Great to see you.
Hey, Matt. Thanks for inviting me. I'd like to thank Deutsche Bank for a great venue, but most importantly, I'd like to thank all the investors for taking some time to hear what we've got to say.
That's great. I think I have a feeling the venue may be a little bit nicer next year in Vegas, but that's 12 months away. Let's maybe get started and as we think about, we've just wrapped up second quarter, we're heading into maybe the second half of the year, and obviously a couple of months left to close out. Dave, maybe you can talk about your top priorities for the organization, before we dig into some more Q&A.
Yeah. Clearly working through the pandemic has been challenging. We quickly pivoted to a remote work environment. We're now in the process to bringing employees back to the office. We've actually gone to mandatory return to all of our U.S. offices, beginning this week. We're going to be going to voluntary return to office, the first week in October for the non-U.S. offices, so that's Canada, across Europe, the U.K., and our Singapore office, and then mandatory in early November. A big priority is bring our employees back safely. Two, we have not rested in terms of upgrading and modifying our network, so we've continued to add capacity, particularly to accommodate the surge in traffic from the increased streaming content. Finally, we're looking to grow our revenues at more historical normal rates.
The pandemic has basically reduced our revenue growth to about one half of our historical average. As businesses return to the office, and I see you're in your office today, Matt, we think that we'll see a re-acceleration in our corporate business and our NetCentric business will continue to perform at or above long-term trend lines.
Yeah, this is not my personal office at home. Maybe in a couple of years, but not yet. Let's start with the corporate business. I know that's a business that historically had been growing double digits for some time. It's give or take, I'd say 60-ish, a little bit over 60% of your revenues. We've seen obviously growth, somewhat pressured the last couple of quarters. Can we talk a little bit about what's been driving some of that weakness, and then maybe more importantly, whether you're seeing any green shoots in the business indicating we may be near an inflection as businesses begin returning to the office?
Yeah. Our corporate business has grown at an average of 11% for the past 15 years organically. In the great financial recession, we had two quarters of negative growth, end of Q4 of 2008, Q1 of 2009, and quickly rebounded. The pandemic has been much longer in duration. We now have five quarters of negative corporate growth. I think there are three reasons for that. One, some of our businesses have shuttered the customers that we have served. Fortunately, that has not been a huge impact on Cogent because we only serve the very best locations, large multi-tenant office buildings in the central business districts of major cities. Most of the companies can afford rents in those types of facilities are generally economically viable, even in a downturn.
Therefore, we haven't seen much increase in business failures, and in fact, our bad debt is below historic averages. The second impact has been on secondary offices, branch locations. Most of our corporate customers need connectivity at their primary location to support their work from home employees. At those secondary locations, they have no employees and no need for bandwidth. We've seen companies turning off those secondary locations. The third impact has been a virtual complete cessation of companies building out private networks to link multiple locations together. Going into the pandemic, we saw a significant amount of our growth in corporate coming from companies who were building either SD-WAN networks or VPLS networks to replace their MPLS legacy internal private networks.
Most companies hit the pause button on those types of programs. Coming out of the pandemic, we're actually seeing a re-acceleration in quotes and companies actually starting to deploy VPNs again. Our corporate business has not recovered to its historical kind of 2.5% sequential growth rate. To be realistic, we're probably still several quarters away from this kind of being in the rear view mirror. We expected to come out of the pandemic maybe a little more quickly in the fall, and the Delta variant has slowed things down. Companies are adding capacity, they are adding incremental locations, and they are beginning to add virtual private networks, but maybe not at the pace they were pre-pandemic. I think it's really a tale of different geographies. A couple of quarters ago, we saw things pretty robust in the South and in the West, and really the Northeast was lagging.
It seems like there's more return to office and acceleration occurring in the Northeast, with the onset of the Delta variant, there seems to be a bit of a pullback, say, in Texas and then in Florida and some of the Western states. It's been a bit rocky, but I do think, the pandemic proved more than anything, businesses need good connectivity. What really differentiates Cogent from many of our competitors is the fact that we sell a non-oversubscribed and non-blocked fiber-delivered symmetric service at prices that are equal or below what our competitors sell an inferior product. For that reason, we continue to feel optimistic that we will be able to return to that 11% corporate growth rate as we're only about 25% penetrated in our footprint, and we have 11% of all multi-tenant office space in North America on net.
Maybe bigger picture is we think about maybe a world where the central hubs, the big multi-tenant office buildings are still around, maybe fully occupied at some point once we come out of this pandemic. There is, I guess, it comes to mind a little bit of maybe headwind around these sort of secondary, tertiary type locations. If we do move, in fact, towards a world where there are these central hubs, but some of these additional locations no longer come back, how would that impact Cogent business?
I think there's two different venues that we need to look at. Roughly 40% of our corporate revenues, 20% of our corporate connections are off-net, meaning we concluded the location is too small for Cogent to build into, and therefore we rent that last mile from another provider. In that segment, we have less than 1% market penetration. While there will be some of those smaller off-net locations that go away, there's such a large addressable market that we don't think it's going to impact us over the long run negatively. The second type of branch office is in another Cogent-lit building, another large multi-tenant building, maybe just in a different city or a different part of the MSA where the main hub office is.
If that is closed down, that actually probably presents an additional opportunity for Cogent, because I agree with what your premise was, Matt, that these buildings are not going to go dark. They're not going to be vacant. There's going to be some other business that moves in to take that space, and therefore it creates an additional customer relationship that we can develop. The good news for Cogent is that when we get in front of a corporate customer, we win between 40% and 50% of the proposals that we issue. That's really an amazing fact considering over 90% of our sales occur over the telephone without the customer ever physically meeting a Cogent sales rep.
Anybody who's been involved in remote sales knows that a 40%-50% close rate on a telesale is truly unheard of, and it really speaks to the value of our network versus what our competitors have to offer.
Let's pivot to NetCentric. That's actually been a business, and by the way, I think at last count, it's a little under 40% of your revenues on a total basis. Big beneficiary of some of the recent events, accelerating growth. Now we're actually seeing that accelerate into double digits the last couple of quarters. From a high level, can you talk about your expectations for the NetCentric business the next couple of quarters, how you're expecting revenue growth to trend? I say this particularly as you're beginning to hit some tougher comps and as you're seeing traffic growth now start to moderate.
Again, I take a long-term view of both segments of our business. Our NetCentric business represents 38% of our revenues, but 96% of our traffic. That business is heavily outside of the U.S., with about 55% of that revenue coming internationally. We today serve approximately 1,300 carrier-neutral data centers in 49 countries around the world. We sell to over 7,500 access networks that are buying their upstream connectivity to the internet from Cogent, and we sell to about 5,000 content-generating customers globally. We have some strong tailwinds to that business. It has averaged a revenue growth rate of 9% year-over-year for the past 16 years. Although, that business tends to be very volatile and very lumpy, with foreign exchange distorting revenue growth, customer concentration, and business model fluctuations, as well as seasonality.
Last quarter, we had the very best NetCentric revenue growth number in the company's history, over 30%, 30.5% year-over-year. Even if you net out FX distortion, it still grew at about 23.5%, or about two and a half times the long-term average. I suspect that business is going to revert back to kind of its average growth rate of about 9% on a revenue basis over the next several quarters, whether it's three quarters, four quarters. Again, it's a little hard to predict with the volatility caused by the pandemic. We do have three really big structural tailwinds that are going to keep that business performing well. The first is demand. It's basically the continued transition of linear video to over-the-top streaming video. Today, almost 40% of video consumed in developed world is streamed. That's up from 18% two years ago.
We clearly saw an inflection up with the pandemic. The second thing that is helping Cogent is the internationalization of that phenomenon. 20 years ago, the internet was predominantly a U.S. phenomenon, with 85% of global traffic in the U.S. Today, the internet is much more globalized, with the U.S. accounting for only about 33% of traffic and continuing to decline. Cogent benefits because of the breadth of our network and the amount of connectivity we have to access networks in over 170 countries. While we have physical network in 50, 49 countries, we have customers who have networks in over 170 countries, and that really benefits us in terms of reach. The third thing that's been occurring at Cogent is the nature of that traffic exchange.
Two years ago, roughly half of our traffic went customer to customer, about half of our traffic was going customer to peer. That has a significant impact on financials because customer to peer, we're only getting paid on one side, customer to customer, we get paid on both sides. Today, we're over 70% on net traffic, meaning it's going from a Cogent content producer to a Cogent access network, we are getting paid on both sides. That has helped our revenue growth. The final factor that helps the total business is the relative mix of traffic by on-net versus off-net, corporate versus NetCentric. In our corporate business, as I stated earlier, roughly 40% of revenues are off-net, 60% of revenues are on-net. When we add $1 of off-net revenue, there's only a 50% gross margin contribution and $0.45 of EBITDA margin contribution.
When we add a dollar of on-net revenue, we get 100% gross margin contribution and $0.95 of EBITDA contribution. In our NetCentric business, 92% of revenues are on-net, with only 8% of revenues being off-net. Even with lower growth rate, we've been able to achieve outsized margin expansion.
Let's maybe pivot a little bit. I know we hit on the two revenue segments, but I want to maybe talk a little bit about the sales force and productivity per rep. I know that was in a little bit of a downward trend to start 2020, but now we've seen thre straight quarters of improvement. Can you maybe help us think about what's driving the sequential pickup we've seen? Then, I guess more broadly, how Cogent can get that back to that 5.1 historical average relative to the 4.5 that we're sitting at as of 2Q?
Yeah. Probably the biggest issue in our sales force was the market. Customers on the corporate side weren't receptive to taking calls. We have seen, probably since last fall, so now for almost a year after the initial shock of the pandemic, customers being a lot more receptive to looking at augmenting and modifying their networks, and therefore receptive to a Cogent sales rep. On the NetCentric side, it's a little bit different, where the customers need what we're selling, and it really comes down to only three questions. Price is number one, and Cogent has always been the market leader, undercutting our competitors by 50%. Two, are you in the location where I need to purchase from you? Three, do you have connectivity that's equal or better than your competitors?
Cogent wins in all of those metrics, and that's why our NetCentric sales have continued strong. The second factor that dampened our sales efficacy was the fact that we had a sales force comprised of 70 teams that sit in 35 Cogent sales offices globally, and they're used to coming in the office every day for eight hours and being on the phone. Almost instantaneously, those individuals had to work from home. We had to equip them with laptops, soft phones. We had to build monitoring systems to manage them remotely. We had to get them acclimated to using our CRM system in a remote environment versus within the office. All of that occurred within a quarter or two of the pandemic, and we took those hits early on in terms of rep productivity. The second thing then had to happen.
Our sales force turns over about 5% a month. That's a big number. It means basically half of our sales force, or a little bit more, is gone each year. We've constantly got to hire and replenish. The nature of the job being an outbound telesales organization is just harder than many people who take the job think it is when they sign up for it. We have some systemic turnover. As a result of that, we had to develop systems to hire and train people remotely. That turned out to be hard. What turned out to be the absolute hardest thing was to actually terminate unproductive people remotely. Today, we have nearly 400 people at Cogent, out of our 1,100 employees, who prior to last week, had never set foot in a Cogent office.
That was really hard, and what was even harder was to hire someone, monitor them, and train them for three or four months, and then term them without ever physically meeting them. We went through a period of modifying our termination procedures. We saw our productivity start to rebound, and I think with employees now back in the office, starting to use systems that we've developed over the past 15 years, we should expect our productivity per rep to continue to increase sequentially this quarter and going forward.
You kind of touched on this, but in terms of sales force turnover, I think historically, you had talked about a 5.2% rate. We had peaked maybe around 7% late last year. We're now back to about 5.6% as of last quarter. We're only 40 basis points higher than, I think, the historical average. In theory, is it fair to assume that with everybody now coming back to the office, that turnover rate should begin to moderate somewhat as well?
I believe so, Matt. I think getting people back in the office where we can help mentor them is both going to make them more productive and make us less likely to look to terminate them. Most of our employee turnover is involuntary. We're a very quota-driven organization with very objective metrics. That's one of the advantages of sales. There's an objective criteria. Did you hit your number or not in a given month? We have really three ways of training a rep. One is classroom instruction, two is online training modules, and we have literally nearly 100 online courses that reps take throughout their career at Cogent. These are one hour modules with different aspects of selling or product education. Maybe the most important tool we have is the third one, which is mentorship, and that means the manager and the regional learning manager.
In our offices, the manager's primary responsibility is to mentor their new hires and make them productive. To supplement the manager, we have 12 regional learning managers who, in normal times, would travel between offices, spending a couple of days in each office a week, holding seminars and classes, answering questions, instructing salespeople on how to overcome customer objections, how to deal with product or technical questions. We were doing that remotely, but the reps were just not as engaged. I think now that we're getting back into the office, we'll see our educational efficacy increase, our rep productivity increase, and therefore, our rep turnover decrease.
Let's pivot to margins. I know in the past, I think you've talked about 200 basis points, and this is a sort of a longer-term vision, but 200 basis points of margin expansion annually, 100 basis points of that coming from gross margins, about another 100 basis points that are OpEx driven. I'm wondering, as we're sitting here with NetCentric maybe punching a little bit above its weight, but corporate lagging, is that margin expansion target something that hinges on getting back to 10% top-line growth? I just want to make sure we're sort of thinking about the longer-term roadmap correctly. That's still intact.
Again, I look at all of our guidance as multi-year and long-term. If we go back in Cogent's history since going public in 2005, we have averaged 10% organic growth in the entire business and just over 200 basis points a year of margin expansion. Because of the pandemic and all of the trends that we've just talked about, we saw our growth rate decelerate to about half of the long-term average, and our margins still improved 130 basis points year-over-year. It didn't quite fall as bad as the revenue growth. Part of that is helped by the mix benefit of better NetCentric and more of the slowdown being on the corporate side, where the contribution margins are not as robust. I think going forward, we expect growth to slowly re-accelerate to historic averages in both businesses, meaning a deceleration in NetCentric, a re-acceleration in corporate.
We also know that our margins will eventually plateau at around 50%. We're at about 39% today. I think that kind of 200 basis points a year guidance for the next five or six years feels correct. It does feel like we're coming out of the pandemic. It does see that we're seeing the green shoots of better funnel building, more customer engagement, lower rep turnover, better rep productivity. All of the leading indicators that show us that our growth is going to re-accelerate are in place, and we are still at 130 basis points of margin expansion with all of these headwinds, and we think that will continue to improve going forward. Again, the rough mix over time is about a 50/50 split in that margin expansion.
About half of it coming from gross improvement because of the operating leverage on our network, about half coming from SG&A efficiencies.
As we think about maybe to tie this all together and bring it to capital allocation, Dave, maybe you can refresh us in terms of how you're thinking about capital allocation priorities between the various buckets, dividend increases, share repurchases, M&A, and maybe we can weave in some CapEx color as well.
Yeah. Those are all of the categories with one other category, which is investing in the business, and there are people in our sales force. The first thing we present to the board is in our quarterly budget reviews, is the fact that we have enough capital to fully invest in our business, meaning we're investing everything that is prudently investable. We're fortunate that we're producing excess cash above that. In terms of CapEx, we have seen our CapEx go up slightly as we've expanded into more international markets.
As data center construction has accelerated, we have continued to grow into those additional data centers, and we'll probably add 100 data centers to our network over the next 12 months, and probably even the next 24 months, 200, just based on what we're seeing in the construction pipeline globally, as well as the likelihood that we'll be adding a few more countries to our footprint. Our multi-tenant office footprint had already matured and is probably only growing at about 3% a year, and it'll probably continue to grow at about that rate, with about 980 million sq ft of multi-tenant office space on net. In terms of CapEx and investing in the business, we're in a pretty good spot. In terms of excess cash, we have returned $1 billion to investors since going public. We're very proud of that.
Virtually all of the new entrants in our industry have not only never returned capital, but have destroyed capital, and the legacy providers, who are typically incumbents, have actually seen perpetual declines in their ability to return capital with declining dividend rates. We feel very comfortable that we have an under-levered balance sheet in a low-interest rate environment, and we are producing excess cash. As a result of that, we have grown our dividend for 37 consecutive quarters sequentially. We do supplement that with buybacks, but with the market at all-time highs, we have erred more over the past year or two to using dividends rather than buybacks to return capital. Investors should expect a methodical growth in free cash flow, a maintenance of leverage at or about our current levels in a low-interest rate environment.
Because of the natural de-levering that occurs, it's kind of a good problem to have. We actually have to constantly look to return more capital, and that's why we have three times accelerated our rate of dividend growth. First it was $0.01, then it went to $0.02 sequentially, now it's $0.025 a share. We think that for the foreseeable future, Cogent will be able to return capital at a growing rate, kind of in line with our cash flow growth rates, which are mid-double digits, mid-teens. Finally, as we think about return of capital, we're always conscious of tax efficiency. As a corporate entity, we have not been a significant cash taxpayer. We don't expect for the next one or two years to be paying any cash taxes.
We have still about $1 billion of NOLs, many of them international, that we'll begin to utilize more effectively. For our shareholders, we have been able to classify over half of our payment in dividends as a return of capital, meaning that it reduces the purchaser's basis in their shares, but they don't pay current taxes and only pay capital gains when or if they sell the Cogent shares. We're trying to be conscious of both our investors' tax situation as well as our own, and returning capital while still investing in the business. We feel we're in a really good place compared to virtually any other company in the telecom ecosystem.
That's great, and thank you for weaving all of that in, Dave. I think we are just about out of time. On behalf of myself, on behalf of everyone at DB, thank you, and we can't wait to see you in person again soon.
I can't wait either. Take care, Matt. Thanks.
Take care. Bye.
Bye-bye.